PDA

View Full Version : Offer: Y-DNA matching with 23andme



geneticexplorer
03-14-2016, 09:50 PM
Recently I've been exploring what the raw 23andme Y-DNA data can offer for insights. They do check for over 2000 positions, many not documented on the ISOGG list. Wonder if the positions selected by 23andme are near perfect matches for R1b-L21 or if it can be used to find recent relatives? Maybe we can find out together?

My offer: if you are on 23andme and have the haplogroup R1b-L21 (R1b1b2a1a2f* in their notation), I will compare your raw Y-DNA information with other R1b-L21 people and report to you privately what positions you differ on.

You can stay completely anonymous. Will only see your raw Y-DNA data, which makes up less than %1 of your DNA.

Send me a PM if you're interested and I'll give you detailed instructions. Your part takes less than 2 minutes.

Also, if you've done anything like this before, I am curious to hear your insights.

Petr
03-14-2016, 11:00 PM
I don't think it will work. I compared 23andMe Y-DNA results of the following persons reported as R1a1a* by 23andMe:
1. R1a1a1a (R-YP942)
2. R1a1a1b1a1a (R-YP1341)
3. R1a1a1b1a2b3 (R-YP237)
and the results (raw data) were exactly the same.

geneticexplorer
03-14-2016, 11:13 PM
Hi Petr, thanks for sharing your findings! It's a bummer the R1a1a1 subclade won't give you more information on 23andMe. :(

The reason I'm more hopeful about R1b-L21, is that I discovered both missing and mismatched mutations (according to ISOGG's data). This could however be due to a 23andMe measurement error or ISOGG's theory still being incomplete. Moreover, I don't know which chip version you have, but with v4, there seems to be quite a bit more Y-DNA data.

Once more R1b-L21 people join, we'll know for sure. :)

ArmandoR1b
03-15-2016, 01:00 AM
23andme tests about 2129 Y-Chromosome regions. They did not test additional Y-chromosome regions in the v4 test since they are interested in medical data and reselling it to medical and pharmaceutical companies.

Out of the 2129 Y-Chromosome regions that are tested only 24 of them are also in the BigTree mutation index at http://www.ytree.net/SNPIndex.php Out of those 24 only 8 of them are subclades of L21. Three of them are now in the ISOGG list one of which was already there in 2008/2009

M222 PAGES00084 R1b1a1a2a1a2c1a1a1a1
L193/S176 R1b1a1a2a1a2c1b1b1a
CTS1751/S3666 R1b1a1a2a1a2c1d
Z17640/Y9130 recurrent in U106
ZS350
ZS351
PF6141
22711460

Most of those SNPs don't have many branches downstream other than M222, which is R1b1b2a1a2f2 at 23andme, and L193 so I don't think many people will walk away with a new subclade by the anonymous report by geneticexplorer. They would be better served by a Yseq L21 panel like the one at http://www.yseq.net/product_info.php?cPath=27&products_id=4321, the FTDNA L21 SNP pack, BigY test, or FGC test.

geneticexplorer
03-15-2016, 02:27 AM
Thanks for the insights ArmandoR1B! Would you say the positions on the BigTree mutation index are the only mutations worth considering? I've found some of my mutations were either associated to other subclades or missing altogether. This made me wonder how my and other people's Y-DNA data match up with each other.

Also, I want to be clear: am not saying that throwing money at the problem, won't help. I am offering this for free. :)

geneticexplorer
03-15-2016, 11:22 AM
Still offering the free service to have your 23andMe Y-DNA analyzed anonymously.
You'll get a list of unique mutations I find including those not on the ISOGG tree.

I'm doing this, because I'm curious to examine the 23andMe R1b-L21 data in more detail.

Send a PM.

geneticexplorer
03-15-2016, 01:23 PM
Some first insights:

- M193 (rs2032676) appears to differ in R1b-L21 people. Have it set to "delete" myself and one other had it set to "insert". How about you?
- Surprising (for me) amounts of "not determined".

Please keep PMing me, look forward to what more we'll find!

ArmandoR1b
03-15-2016, 01:48 PM
Thanks for the insights ArmandoR1B! Would you say the positions on the BigTree mutation index are the only mutations worth considering?
If they aren't on the BigTree mutation index there is an extremely low probability of finding a L21 SNP that can be found in a significant number of people.


I've found some of my mutations were either associated to other subclades or missing altogether. This made me wonder how my and other people's Y-DNA data match up with each other.
Mutations associated to other subclades are called recurrent SNPs. Z17640 Y9130 rs1946603 9850346-A-G is one of those because it is found in U106 and L21. It's only going to be listed if enough people have that mutation. Otherwise, it can be considered a private mutation, or even a false positive, and therefore you'll be lucky to find others with the same mutations. If you do find other people with the same different mutation than other L21 people it does not mean that they are in the same subclade. The mutations can be in completely different subclades of L21. I suppose the one you are calling missing are no-calls which only additional testing can determine the result.


Also, I want to be clear: am not saying that throwing money at the problem, won't help. I am offering this for free. :)

The free service is extremely limited and useful to a less than 0.1% of L21 people, other than the SNPs found in the BigTree mutation index, and can't put people on important branches like the paid services can.

In order to make the searching easier for the ones I mentioned in a previous post I included the RSID that can be searched at https://www.23andme.com/you/explorer/ or the raw data file can be searched which will provide whether or not the person is ancestral or derived.


SNP rsid hg19 POSITION-ANC-DER ISOGG 2016 Longhand (March 14 2106)
M222 PAGES00084 rs20321 14902414-G-A R1b1a1a2a1a2c1a1a1a1
L193 S176 i5049637 6753344-G-A R1b1a1a2a1a2c1b1b1a
CTS1751 S3666 rs11799049 14048611-A-G R1b1a1a2a1a2c1d
Z17640 Y9130 rs1946603 9850346-A-G recurrent in U106
ZS350 rs13305055 9131450-G-A same branch as ZS351
ZS351 rs13303690 9131459-C-G same branch as ZS350
PF6141 rs9786197 21117888-C-T
noname rs9306842 22711460-T-TA

ArmandoR1b
03-15-2016, 01:59 PM
- M193 (rs2032676) appears to differ in R1b-L21 people. Have it set to "delete" myself and one other had it set to "insert". How about you?
M193 (rs2032676) is from haplogroup T and is position 15014505. http://www.isogg.org/tree/ISOGG_YDNA_SNP_Index.html If a lot of L21 people had it to set to "insert" then it would already be in the BigTree index at http://www.ytree.net/SNPIndex.php If you do find others with it set to "insert" they aren't necessarily in the same L21 subclade. More SNP testing would be necessary to determine if they are in the same L21 subclade or not.


- Surprising (for me) amounts of "not determined".
No-calls are very common in the 23andme Y-chromosome test. That hasn't changed since 2008. http://archiver.rootsweb.ancestry.com/th/read/GENEALOGY-DNA/2008-10/1223425467

geneticexplorer
03-15-2016, 03:30 PM
Armando, thanks again for your insightful comments and hints on the terminology.

Some clarifications:

1. There appear to be at least two different ways Y-DNA is approached. One approach is starting from Y-Adam and building a haplogroup tree (top down). The approach I'm trying (bottom up), is starting from the pool of autosomal matches I get (from e.g. 23andme), and use the Y-DNA data to see if we have a common ancestor on the paternal line. Wonder if private mutations (like the one mentioned previously) are helpful in that context or if the mutations might just as well be false positives.

2. Checked the list you provided. Unless it's a false positive, the ZS350 and possibly 22711460 mutation seem to apply to me.

M222 - negative
L193 - no call
CTS1751 - negative
Z17640 - no call
ZS350 - positive
ZS351 - no call
PF6141 - negative
22711460 - positive* (I have a T, not a TA)

reader
03-15-2016, 03:57 PM
A couple thousand individuals have shared 23andMe (or in some cases FTDNA or Ancestry) raw data here:

https://opensnp.org/

This can be freely downloaded and should likely include hundreds of L21+ individuals.

Some people have also uploaded 23andMe data here:

https://my.pgp-hms.org/public_genetic_data

ArmandoR1b
03-15-2016, 03:59 PM
1. There appear to be at least two different ways Y-DNA is approached. One approach is starting from Y-Adam and building a haplogroup tree (top down). The approach I'm trying (bottom up), is starting from the pool of autosomal matches I get (from e.g. 23andme), and use the Y-DNA data to see if we have a common ancestor on the paternal line. The Y-DNA tree is built by comparing a large pool of people that have been tested. It's logically concluded that the SNPs shared by everyone were passed down from Y-Adam. From there the SNPs that are shared by a smaller group is a different branch. They keep comparing to find the branches.

By comparing a pool of autosomal matches you are comparing many people that are very likely from your same ethnic group but very distantly related in the direct paternal line when they are also L21, because most aren't from your direct paternal line. The minute amount of people that you find with similar mutations aren't going to help you find people that are related to you in the direct paternal line in a genealogical timeframe.


Wonder if private mutations (like the one mentioned previously) are helpful in that context or if the mutations might just as well be false positives.The odds are not in your favor of finding a mutation tested by 23andme that is genealogically relevant.


2. Checked the list you provided. Unless it's a false positive, the ZS350 and possibly 22711460 mutation seem to apply to me.

M222 - negative
L193 - no call
CTS1751 - negative
Z17640 - no call
ZS350 - positive
ZS351 - no call
PF6141 - negative
22711460 - positive* (I have a T, not a TA)
If you have an A for ZS350 then you need to test positive for the following in order for that SNP to be relevant. DF13 > L513/S215/DF1 > S5668 > Z16357 > Z16343 > Z16854 and none of those are tested by 23andme.

To be positive for 22711460 you need to have TA and for it to be relevant you have to be positive for the following DF13 > CTS1751 > ZZ49 > Z17966 > BY596 > BY595 and none of those are tested by 23andme.

geneticexplorer
03-15-2016, 04:49 PM
Hi reader, thanks for stepping out of your reader role and sharing those links! They are super useful to me! :)

Was looking on Genbank also, but only found lots of mtDNA samples so far.

vettor
03-16-2016, 06:00 AM
M193 (rs2032676) is from haplogroup T and is position 15014505. http://www.isogg.org/tree/ISOGG_YDNA_SNP_Index.html If a lot of L21 people had it to set to "insert" then it would already be in the BigTree index at http://www.ytree.net/SNPIndex.php If you do find others with it set to "insert" they aren't necessarily in the same L21 subclade. More SNP testing would be necessary to determine if they are in the same L21 subclade or not.




What does this mean in the mutation of M193?

del->AACA


BTW , as a T .........I am negative for M193

geneticexplorer
03-16-2016, 04:53 PM
vettor: I think it means "nothing" gets replaced with the sequence "AACA". See https://en.wikipedia.org/wiki/Insertion_(genetics)

geneticexplorer
03-16-2016, 04:57 PM
Everyone: I finished the analysis for now.

Conclusion: Comparing Y-DNA for R1b-L21 on 23andMe (for now) appears to not give major insights. Found many perfect Y-DNA matches and some with only minor differences. Bummer!

ArmandoR1b
03-16-2016, 06:43 PM
Conclusion: Comparing Y-DNA for R1b-L21 on 23andMe (for now) appears to not give major insights. Found many perfect Y-DNA matches and some with only minor differences. Bummer!
That's exactly what Petr and I had been trying to explain would happen. It is also why a Yseq L21 panel like the one at http://www.yseq.net/product_info.php...oducts_id=4321, the FTDNA L21 SNP pack, BigY test, or FGC test is needed to determine which branch an L21 person belongs in.

geneticexplorer
03-16-2016, 06:55 PM
Armando: I'm glad several of us got the same conclusion! :)

You mentioned L21 products. Am assuming you tried some of them yourself. What were the most important insights you gained for your family research?

Mikewww
03-16-2016, 08:11 PM
Some first insights:

- M193 (rs2032676) appears to differ in R1b-L21 people. Have it set to "delete" myself and one other had it set to "insert". How about you?
- Surprising (for me) amounts of "not determined".

Please keep PMing me, look forward to what more we'll find!
I saw mention of L193 earlier. I just want to make sure there is no confusion on this. M193 has nothing to do with R1b-L21 subclade people. L193 does. L193 marks a subclade under R1b-L21 in a big branch of L21 marked by L513.

ArmandoR1b
03-16-2016, 08:15 PM
I saw mention of L193 earlier. I just want to make sure there is no confusion on this. M193 has nothing to do with R1b-L21 subclade people. That was covered in post #9


L193 does. L193 marks a subclade under R1b-L21 in a big branch of L21 marked by L513.

That is why I had listed it in post #8

ArmandoR1b
03-16-2016, 08:26 PM
Armando: I'm glad several of us got the same conclusion! :)Yeah, sometimes people have to see it for themselves even after it has been explained.


You mentioned L21 products. Am assuming you tried some of them yourself. What were the most important insights you gained for your family research?

I'm not L21. I have had a lot of family members get similar tests for other haplogroups and therefore I know what they have to offer. BigY or FGC are the best tests. You can't get as much information from the other testing but you have to be willing to spend the money. The Yseq L21 panel will put you into a subclade if you are positive for one of the SNPs they test but it won't test anywhere close to the number of SNPs that the BigY or FGC tests.